Battery State Estimation Using Voltage-Current Gradient Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional neural networks struggle to accurately estimate the state of charge (SOC) and deterioration degree (SOH) of secondary batteries with varying electrical characteristics from different manufacturers and models, limiting their applicability across diverse battery types.
Innovation Solution
A learning method using machine learning to estimate SOC and SOH, which involves preprocessing terminal current and voltage data to calculate differences and gradients, incorporating open circuit voltage and gradient change rates, and employing a state estimation model configured with RNN, LSTM, or CNN to accurately predict battery states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional neural networks are trained with measured values of voltage, current, and internal impedance to estimate SOC and SOH, then estimation accuracy for a specific manufacturer and model is improved, but adaptability to batteries from different manufacturers and models deteriorates
Solution Approach 1:
The patent transforms the input parameters from direct measured values (voltage, current, internal impedance) to derived parameters representing electrical characteristics (gradient of voltage with respect to current, higher-order gradients). This parameter transformation enables the neural network to learn universal relationships across different battery types while maintaining estimation accuracy, resolving the contradiction between precision and adaptability
Solution Approach 2:
The patent segments the estimation process into two stages: first extracting electrical characteristics through gradient calculations, then using these extracted features as inputs to the neural network. This segmentation allows the network to focus on learning the relationship between electrical characteristics and battery state, rather than being confounded by manufacturer-specific variations in raw measurements
2Measurement precision
If one neural network is trained for a specific manufacturer and model to achieve accurate estimation, then estimation accuracy is improved, but device complexity increases due to needing multiple networks for different battery types
Solution Approach 1:
The patent creates a universal neural network that can accurately estimate SOC and SOH for multiple manufacturers and models simultaneously. By using transformed parameters (gradients of voltage with respect to current) as inputs, the single network achieves multi-functionality, eliminating the need to maintain separate networks for different battery types and thus reducing device complexity
Data Source
AI summary
A learning method of a state estimation model of a secondary battery includes training the state estimation model to learn a relationship of a state estimation input data preprocessed from state variables including measured terminal currents and terminal voltages of the secondary battery with a charge rate or a deterioration degree of the secondary battery. The state estimation input data includes time-series data of difference gradients, which is a change rate of the differences of the terminal voltages with respect to the differences of the terminal currents.